PostgreSQL 10 introduces table partitioning, which is a significant improvement for many use cases over the much older table inheritance feature. Rather than focusing on the positive, in this talk I'll talk about every known problem with the partitioning feature: performance problems (such as the fact that we still rely on constraint exclusion for partition pruning), usability problems (such as painful locking requirements), and missing features (such as hash partitioning). Some of the problems addressed in this talk will likely be addressed in PostgreSQL 11; others will not. All of these problems represent opportunities to make PostgreSQL's partitioning feature even better in the future.
The following slides have been made available for this session: